recursive-context-pruning-token-budgeting

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is purely instructional and provides a methodology for the AI agent to manage its context window and response style. It does not involve any code execution, file system access, or network communication.
  • [SAFE]: Includes defensive instructions specifically designed to maintain safety headers and system-level constraints during context compression to prevent potential jailbreaks.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 04:10 AM